Why DefiLlama's TVL Differs from Project Official Sites

On-Chain

28 July 2026

The TVL shown on DefiLlama often differs from a project's own website; this is normal, not a bug – their scopes of measurement, data sources, and even what counts as 'TVL' are different. The discrepancy can arise from differing statistical scope, data update time lags, or even active data corrections made by DefiLlama.

Reason 1: Different Statistical Scope – DefiLlama Counts 'Core TVL,' Projects May Count Everything

This is the core source of discrepancies. DefiLlama has its own criteria for what constitutes TVL, which may not align with the numbers a project wants to present.

Staking TVL Excluded from Core TVL

Many DeFi protocols, besides their core business (e.g., lending, trading), have a staking contract where users lock protocol governance tokens (like CRV, AAVE) to earn rewards. DefiLlama explicitly excludes staking TVL from core TVL for two reasons: first, to avoid penalizing protocols that don't have a token or staking program; second, staking TVL is easily manipulated—projects can offset losses by minting more governance tokens, so its liquidity risk perception is less acute than core activities.

If a project's official website shows 'total TVL' (core + staking) while DefiLlama only shows core TVL, the two naturally won't match.

Certain Nodes Not Included in Chain TVL

When calculating the total TVL of a chain, DefiLlama excludes native asset staking (e.g., staking ETH on Ethereum, ADA on Cardano). The reasoning: if included, the chain's TVL would basically mirror the market cap of its native token, overwhelming real DeFi protocol usage. Cardano's on-chain TVL is about $330 million, but staked ADA is worth around $15.8 billion—if DefiLlama included the latter, DeFi's share would drop to 2%, obscuring actual usage data.

Reason 2: Data Update Time Lags – Page Caching, API Refreshes First

DefiLlama updates data hourly—TVL, total borrow, treasuries, stablecoin supply, etc., all refresh every hour. However, the web front end may lag behind the API by up to one hour due to caching. If a project's site updates in real time while DefiLlama is showing an hour-old snapshot, the figures will differ.

Reason 3: DefiLlama May Proactively Conduct 'Due Diligence' on Data – Removing What Looks Inflated

This is a professional practice of data platforms, and the most critical variable causing data gaps.

Example: Figure TVL Controversy (September 2025)

Figure claimed its RWA on-chain assets reached $12 billion, but DefiLlama's founder publicly responded: 'Assets that cannot be verified on-chain are not counted as TVL.' Several red flags were detailed:

On-chain verifiable assets: ~$5 million in BTC, ~$4 million in ETH, drastically out of line with the claimed $12 billion.
Self-issued stablecoin supply only $20 million, far too little to support large-scale transactions.
Most asset transfers were executed by other accounts, with addresses showing almost no on-chain interaction, suggesting an internal database mirror rather than real transactions.

If a project's website treats 'book assets' from its internal database as TVL, while DefiLlama only recognizes on-chain verifiable deposits, the gap can span several orders of magnitude.

Reason 4: Algorithm Differences – Weighting Formulas Vary

Because of DeFi's composability ('lego' nature) – a deposit in one protocol may be re-deposited as an LP token in another – TVL suffers from double counting. DefiLlama offers a 'Doublecount' toggle so users can decide whether to include such re-used assets.

Different data platforms apply different weighting to protocol layers. For example, first-layer (real stablecoins in 3Pool) might have a weight of x1, second-layer (Metapool using 3Pool LP tokens) might be x1/2. If a project's own algorithm stacks all layers, and DefiLlama uses a different weighting, the results won't match.

How to Decide Which Number Is More Reliable

Data transparency: DefiLlama's adapter code is open-source and subject to review. Project websites rarely disclose detailed TVL calculation logic; their stated numbers can sometimes be a 'marketing metric.'
Verifiability: On-chain data is public. If you can verify through Etherscan that a particular asset actually resides in the protocol's contract, then it qualifies as 'true TVL.' DefiLlama's core principle is: count only TVL verifiable directly on-chain. If a figure cannot be verified on-chain, it will not be included on such data platforms.

How to Determine Which Number You Should Look At

First, clarify why you're checking TVL. If you're making an investment decision or evaluating actual protocol usage, prioritize DefiLlama's data because its methodology is standardized, comparable, and open-source reviewed. If you're just looking at 'marketing-facing' numbers for a project, the official site figure can serve as a reference – but understand that the measurement scope is likely different, and the official number is probably a 'broad scope' version.